13 replies to this topic

[kconf]

Hey all,

I was wondering if there is any guidance on how many lots can be composited when sending samples to lab for micro analysis. We are "no water" facility. The tanks are never washed. Does the batch size matter? 

[MDaleDDF]

finished product testing?

Edited by MDaleDDF, 16 March 2026 - 02:47 PM.

[kconf]

Yes finished RTE. We are a chocolate manufacturing facility. 

[kconf]

Low risk

[MDaleDDF]

Hmm.   I have never heard of a number you can't surpass, but the obvious issue is if you have a hit, you're now recalling however many lots you made.   But if you're low risk and you're comfortable with your recall history, etc, go for broke.   But it sure would make me nervous.   All it takes is one supplier to call and say they found something on their end that triggers a recall, and you're screwed... (this has literally happened to me in my time here)

[kconf]

Is 5 too many? My point is that if one comes back positive, the chances are last 3-4 need investigation as well since there is always some residue left in tanks. Our SOP says 3 max. 

[Scampi]

The size is dependent on your companies risk comfort and past sampling history

 

why are you wanting to composite and not do the testing by batch?

[kconf]

Because it literally is the same thing. 

 

Sugar - Lot XYZ

Liquor - Lot 123

Butter - Lot 567

 

Same ingredients with the same lot #. Just different days. Is it still unreasonable to composite multiple lots? 

[jfrey123]

As stated it above, the issue becomes whenever you have a hit on the composited material.  You can't pick and choose which lots are affected by the potential positive, so at minimum then you're facing a hold or recall of however many lots were in the composite.  I've seen companies treat composite positive as an indicator only that triggers mandatory testing of individual lot samples afterward (which I don't love because they'll tend to find the follow up tests are negative and they just move on).

 

I understand your point about being the same lots of input ingredients, so a composite of all finished lots which used the same inputs does make sense.  I'm not sure I love it given the multiple day differences: 3 batches made Monday, 2 batches made Tuesday, all composited into one sample for testing?  Unless it's running 24 hours a day for those periods, I'd want the samples pulled from different days to account for any human error factors.

 

For my own morbid curiosity: how do you get away with never cleaning your tanks?

[kconf]

We have never cleaned tanks in years. If we introduced water, we would be in trouble. There is no water in our formulation either. We just add ingredients, mix, and keep on going. 

 

We typically make one batch a day. If it is large, it may run 2-4 days. Then a lot number is changed when the tank is ~empty. New ingredients are added, new lot number is assigned for this new batch. This can happen 3-5 times a week. If this happened 3 times a week, 3 samples are composited into one. 

[MDaleDDF]

Does every lot get tested via your combining them?    Or quarterly testing, or what?    How much is leftover in your tank when it's emptied?   This sounds like a silo situation, so you'd be going backward and forward in the event of a recall anyway, no?

 

I don't know anything about the chocolate biz but it seems odd to me your tanks aren't getting cleaned on some sort of schedule justified by a risk analysis.   Are you GFSI certified?   I believe from your previous posts you are?

Edited by MDaleDDF, 16 March 2026 - 04:44 PM.

[kconf]

Yes GFSI. We have never had any issues (with not washing). Chocolate and water do not mix. When dirty, it just needs to be scraped. You can go centuries without ever needing to introduce water to tanks for cleaning. We are super clean! 

 

Every lot gets tested. The composition varies. It could be one lot or up to three lots. We send samples weekly. 

 

Tanks may have anywhere from 5-20 lbs. It is insignificant when a batch is 5k-20k lb. 

[MattQA]

https://ahrouj-andre...t_sampling_app/

 

Here's a tool that simulates testing outcomes for different sampling plans. For example, if you make 3500 bars a day and have an intermittent issue that contaminated 500 of them and you composited 5 samples, your chance at catching a positive is about 50/50. Same scenario, but with 20 samples, your chances go up to 95%. Now, 20 samples is a lot to composite so that should be broken up because the lab will take a small portion of that to test. It might not be detected if the bacterial load is light.

 

It illustrates how we can miss an issue without the proper sampling plan. It might also explain how companies can start with a limited recall that expands to a massive recall - other lots were tested negative but the sampling plan wasn't good enough to catch the intermittent issue.

 

As far as guidance goes, SQF suggests you take FP samples 3-4 hours after production start.

[MDaleDDF]

Well if you're GFSI inspected obviously risks have been assessed and dealt with.

So back to your original question:    I would think if you're background sampling is clean enough you could pretty much go with whatever number you're comfortable with, within reason, as long as management understands the risk if something should go wrong.  Just update your SOP to a new REV that says 5 max, and rock and roll....

Edited by MDaleDDF, 16 March 2026 - 04:57 PM.